What this script does:
- proceeds from the output of preprocessing script.
- takes as input data_processed_step1.csv, and data_model_prediction.csv - plots Sorting and Causal Task rdms, gets correlations between empirical and models RDM, gets within subject correlations.
- outputs data_freesort_mins_causals.csv for use in study 2 script for item selection.

1 Setup

1.1 load packages

1.2 Set variables

1.3 Read data

1.4 Reshape data

Create separate data frames for Sorting Task, Causal Task and a combined data frame

2 Freesort RDMs

2.1 plot subject-specific raw responses

2.2 plot average raw responses

2.3 plot subject-specific Sorting Task rdms

2.4 figure for paper: plot average freesort rdm

3 Causal RDMs

3.1 plot subject-specific causal RDMs

3.2 figure for paper: plot average causal RDM

the tris show a mixture of what you are causing and what is being caused. the more top right you are, the more you’re showing what you cause, and the lower you are, the more you’re showing being caused top half of items: your the cause mostly, middle item: equal likelihood, bottom half: your ethe effect mostly

3.3 Specific causal relations for paper

## mind → action: mean distance = 0.323 [95% CI: 0.309, 0.337]
## action → mind: mean distance = 0.415 [95% CI: 0.399, 0.431]
## action → action: mean distance = 0.611 [95% CI: 0.592, 0.630]
## body → body: mean distance = 0.381 [95% CI: 0.362, 0.400]
## body → action: mean distance = 0.442 [95% CI: 0.424, 0.460]
## action → body: mean distance = 0.427 [95% CI: 0.410, 0.445]
## perception → cognition: mean distance = 0.149 [95% CI: 0.130, 0.167]
## cognition → perception: mean distance = 0.477 [95% CI: 0.442, 0.512]
## seeing → sick: mean distance = 0.390 [95% CI: 0.300, 0.481]
## hearing → sick: mean distance = 0.606 [95% CI: 0.514, 0.697]

4 Correlations: Sorting vs Causal Task

4.1 mean within-subject correlation

## [1] 0.03372421

4.2 t-test

## 
##  One Sample t-test
## 
## data:  subject_correlations
## t = 3.9823, df = 49, p-value = 0.0002258
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
##  0.01670612 0.05074230
## sample estimates:
##  mean of x 
## 0.03372421

4.3 t-test assumptions

seem fine, so no need for perm test

## 
##  Shapiro-Wilk normality test
## 
## data:  subject_correlations
## W = 0.95551, p-value = 0.05768

5 Correlations: Data vs Models

5.1 General functions for correlation plot

5.2 noise ceiling functions

5.3 Calculate Sorting Task noise ceiling

## $upper_bound
## [1] 0.2596685
## 
## $lower_bound
## [1] 0.2595976

5.4 Calculate Causal Task noise ceiling

## $upper_tri_upper_bound
## [1] 0.4409539
## 
## $upper_tri_lower_bound
## [1] 0.4413385
## 
## $lower_tri_upper_bound
## [1] 0.4124777
## 
## $lower_tri_lower_bound
## [1] 0.4130283

5.5 figure: plot correlations both tasks

5.6 t-test

## # A tibble: 8 × 5
##   method       correlation_type  mean_mean_correlation t_value  p_value
##   <fct>        <chr>                             <dbl>   <dbl>    <dbl>
## 1 Sorting Task 2 Category                       0.144     5.99 2.39e- 7
## 2 Sorting Task 3 Category                       0.254     9.76 4.51e-13
## 3 Sorting Task 6 Category                       0.194    11.1  5.77e-15
## 4 Sorting Task Cosine Similarity                0.130     7.95 2.29e-10
## 5 Causal Task  2 Category                      -0.0823   -4.96 8.90e- 6
## 6 Causal Task  3 Category                      -0.0357   -2.93 5.10e- 3
## 7 Causal Task  6 Category                      -0.0569   -6.20 1.16e- 7
## 8 Causal Task  Cosine Similarity               -0.0797   -7.49 1.14e- 9

6 Freesort minus causal distances for Study 2

Here we are geenerating freesort minus causal distances to itemB (i.e. the target item). These will be used to select items for study 2 in the study 2 script.

7 Order effects

7.1 causal-causal vs freesort-freesort correlation

## [1] "Corr(causal rdm Causal Task first, causal rdm Sorting Task first): 0.951633021558085"
## [1] "Corr(freesort rdm Causal Task first, freesort rdm Sorting Task first):  0.861761299764209"
## [1] "correlation difference: 0.0898717217938758"

7.2 causal-freesort vs causal freesort correlation

## [1] -0.05419336
## [1] "Corr(causal rdm Causal Task first, freesort rdm Causal Task first): -0.0541933616487223"
## [1] "Corr(causal rdm Sorting Task first, freesort rdm Sorting Task first):  0.0304316889389261"
## [1] 0.08462505
## [1] "Correlation difference: 0.0846250505876484"
## [1] "Z-score: 0.207441401412369"
## [1] "P-value: 0.83566515206997"